As DRAMs increase in memory cell density, there is a continuing challenge to maintain sufficiently high storage capacitance despite decreasing cell area. Additionally, there is a continuing goal to further decrease cell area. One principal way of increasing cell capacitance is through cell structure techniques. Such techniques include three-dimensional cell capacitors, such as trenched or stacked capacitors. Yet as feature size continues to become smaller and smaller, development of improved materials for cell dielectrics as well as the cell structure are important. The feature size of 256 Mb DRAMs will be on the order of 0.25 micron, and conventional dielectrics such as SiO.sub.2 and Si.sub.3 N.sub.4 might not be suitable because of small dielectric constants.
Highly integrated memory devices, such as 256 Mbit DRAMs, are expected to require a very thin dielectric film for the 3-dimensional capacitor of cylindrically stacked or trench structures. To meet this requirement, the capacitor dielectric film thickness will be below 2.5 nm of SiO.sub.2 equivalent thickness.
Insulating inorganic metal oxide materials, such as ferroelectric materials or perovskite material or pentoxides such as tantalum pentoxide, have high dielectric constants and low leakage current which make them attractive as cell dielectric materials for high density DRAMs and non-volatile memories. Perovskite material and other ferroelectric materials exhibit a number of unique and interesting properties. One such property of a ferroelectric material is that it possesses a spontaneous polarization that can be reversed by an applied electric field. Specifically, these materials have a characteristic temperature, commonly referred to as the transition temperature, at which the material makes a structural phase change from a polar phase (ferroelectric) to a non-polar phase, typically called the paraelectric phase.
Despite the advantages of high dielectric constants and low leakage, insulating inorganic metal oxide materials suffer from many drawbacks. For example, all of these materials incorporate oxygen or are otherwise exposed to oxygen for densification to produce the desired capacitor dielectric layer. Unfortunately, the provision of such layers or subjecting such layers to oxidation densification also undesirably oxidizes the underlying bottom or lower storage node plate, which is typically conductively doped polysilicon. For example, Ta.sub.2 O.sub.5 is typically subjected to an anneal in the presence of an oxygen ambient. The anneal drives any carbon present out of the layer and advantageously injects additional oxygen into the layer such that the layer uniformly approaches a stoichiometry of five oxygen atoms for every two tantalum atoms. The oxygen anneal is commonly conducted at a temperature of from about 400.degree. C. to about 1000.degree. C. utilizing one or more of O.sub.3, N.sub.2 O and O.sub.2. The oxygen containing gas is typically flowed through a reactor at a rate of from about 0.5 slm to about 10 slm.
The art accepted solution to avoiding oxidation of underlying silicon is to provide an intervening oxidation barrier layer between the underlying conductive polysilicon and overlying insulating inorganic metal oxide dielectric layer. This layer is accordingly desirably electrically conductive, as the underlying polysilicon will be in electrical connection with the insulating inorganic metal oxide dielectric layer.
There are a limited number of oxidation barrier materials which are conductive. Elemental platinum (a conductive oxidation barrier) on polysilicon has been suggested as a composite layer or construction for a lower capacitor plate, but undergoes physical degradation with thermal cycling due to silicon diffusion through the platinum. Sputtered TiN and CVD TiN (other conductive barriers) have been known to fail due to diffusion along grain boundaries.
Further even with oxidation barriers, some oxidation of the underlying silicon can and does occur the more elevated the oxidation anneal temperature. High temperature anneals in non-oxidizing atmospheres have been used as a substitute for the oxygen anneal. Such have the advantage of achieving or repairing crystal structure without oxidizing the underlying silicon. However, the lack of oxygen prevents significant densification and homogenous production of the stoichiometric oxide. Thus, less than desirable dielectric constant will typically be achieved.